jasongrossman has quit [Ping timeout: 245 seconds]
jasongrossman has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.3]
Taneb has quit [Quit: I seem to have stopped.]
pie___ has joined #nixos-chat
pie__ has quit [Ping timeout: 245 seconds]
jasongrossman has quit [Ping timeout: 246 seconds]
jasongrossman has joined #nixos-chat
endformationage has quit [Quit: WeeChat 2.3]
asymmetric has joined #nixos-chat
aszlig has quit [Quit: Kerneling down for reboot NOW.]
aszlig has joined #nixos-chat
asymmetric has quit [Ping timeout: 246 seconds]
johanot has joined #nixos-chat
<etu>
I find a bit weird and funny that *stable* is 5.0.1 and *mainline* is 5.0. So stable is newer than mainline? https://www.kernel.org/
__monty__ has joined #nixos-chat
<andi->
There hasn't been a RC0/1 yet
<andi->
so that is natural
<andi->
stable is bugfixing things that might happen in the mainline tree. Those fixes in 5.0.1 might be queued up for 5.1 already but not yet in a release.
Taneb has joined #nixos-chat
jtojnar has joined #nixos-chat
Jackneill has quit [Quit: Leaving]
Jackneill has joined #nixos-chat
jasongrossman has quit [Ping timeout: 246 seconds]
<infinisil>
Aw man, I just realized
<infinisil>
For this on-demand live-streaming from my laptop, I set up all this systemd socket crap
<infinisil>
And I ended up using `ssh <machine> ncat -U /run/cam/socket | mpv -` to view the stream (ncat -U just connects to a socket)
<infinisil>
Well the socket starts a service, and the service just runs a command..
<infinisil>
So that's stupid.. I could just do `ssh <machine> <the command> | mpv -`..
<infinisil>
No need for any sockets or services
<andi->
It's what I keep telling people that using NC (or ssh) togeter with ffmpeg/mpv is an awesome teleconferencing system ;) No need for fancy CPU burning browsers..
<infinisil>
andi-: Oh neat, do you have some ready-to-go commands for that?
<andi->
infinisil: I do not since nobody was (in)sane enough to do that with me yet.. And conferencing with myself is kinda stupid :)
<infinisil>
Hehe yeah
<andi->
I build some IPTV streaming platform on a past life and thus I think ffmpeg might be the right tool for the job.
<infinisil>
I'd be the kind of guy to want to do this instead of ugly skype or so
<etu>
andi-: I see, that makes sense
<infinisil>
andi-: I wrote this here https://github.com/Infinisil/picstream , which allows you to stream a slideshow to any number of clients, you control when the clients see the next picture
<infinisil>
And it's just a couple lines of bash!
<andi->
infinisil: nice, you could probably abuse "motion jpeg" which is fancy speech for concatinated jpegs
<andi->
and just use one socket for all of it
<infinisil>
Huh how would that work
<andi->
just send pictures one after the other with the right mime/type/codec/… setup on the receiving side
<infinisil>
I see, a good thing about the current solution though is that it works with a standard feh viewer, so you got all the features it has for viewing already. And it requires only very minimal dependencies (python, nmap, and feh)
<infinisil>
I guess a browser would be able to view it then, but I'd rather not involve any browser
<andi->
mpv is also fine :)
<andi->
but yeah there are many simpler solutions for things like sharing picture. If you want to have it in perfect sync and Many-to-Many then those solutions all come to and end quickly.
jasongrossman has joined #nixos-chat
<infinisil>
I am planning on making a new version of the program, written in haskell, with features like allowing many clients to connect, view and queue pictures. Also maybe deduplicating them automatically, only showing ones not everybody has seen already (using something like https://github.com/Infinisil/soph)
jasongrossman has quit [Quit: ERC (IRC client for Emacs 26.1)]
<elvishjerricco>
Hm. It takes much longer to sync data in 8M increments to my external HDD than it does with e.g. 4G increments. I would have thought that 8M was large enough to defeat the small sync issues. Turns out this is the cause of all my Time Machine performance woes; Time Machine only writes in 8M blobs, and it does so in bursts with down time in between.
<gchristensen>
ouch
<mdash>
elvishjerricco: it was probably fast on some hardare 10 years ago
<mdash>
:-/
drakonis has joined #nixos-chat
<elvishjerricco>
Yea... I really want to abandon Time Machine, but I dunno if there's anything that lets you restore a whole system, including settings, apps, and everything else, that works for macOS.
<elvishjerricco>
mdash: Oh that's interesting. Though I'm guessing reading one byte from the sparse bundle that way probably requires reading a gigabyte or two of bands :/
<mdash>
likely
<elvishjerricco>
I don't really know how HFS works at all; will it result in fragmented bands such that there's tons of mostly worthless 1GB files? You'd probably need to make HFS think the block size = the band size somehow
<gchristensen>
TM is the easiest, most low-friction way
<gchristensen>
and is pretty fast once your initial backup is done
drakonis1 has joined #nixos-chat
yl has joined #nixos-chat
johanot has quit [Quit: WeeChat 2.2]
tilpner_ has joined #nixos-chat
pie___ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
tilpner has quit [Ping timeout: 268 seconds]
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
drakonis1 has quit [Quit: WeeChat 2.3]
drakonis_ has joined #nixos-chat
asymmetric has joined #nixos-chat
drakonis has quit [Ping timeout: 245 seconds]
drakonis_ has quit [Read error: Connection reset by peer]
drakonis_ has joined #nixos-chat
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 252 seconds]
asymmetric has quit [Ping timeout: 240 seconds]
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 252 seconds]
drakonis has joined #nixos-chat
drakonis_ has quit [Ping timeout: 252 seconds]
<elvishjerricco>
gchristensen: Yea. Somehow it's going fast all of a sudden, and all I changed was using an ip instead of the avahi-advertised service, so maybe I'll keep with it
<gchristensen>
weird
jD91mZM2 has quit [Ping timeout: 246 seconds]
<elvishjerricco>
I'll probably still do something like znapzend on the volume it backs up to though. HFS doesn't inspire nearly the same degree of confidence that ZFS does, so it'll be good to be able to roll the filesystem's internal state back in case macOS screws it up somehow
<mdash>
"ZnapZend is written in modern object oriented Perl"
<mdash>
OK I'm thinking twice about that
drakonis_ has joined #nixos-chat
<infinisil>
mdash: I've been using znapzend for my backups for a while
<infinisil>
I can moderately recommend it
drakonis has quit [Ping timeout: 240 seconds]
<infinisil>
It works generally, but I sometimes had some problems
<infinisil>
I think a bug I've encountered has since been fixed though
<gchristensen>
any germans available to PM?
<lassulus>
uhm yes?
<cransom>
re: dkorak. well, after a little over a week of hair pulling anxiety, I'm at least back to 30 wpm-ish. At least at this level I'm not swearing constantly about not being able to get the thoughts out fast enough.
<tilpner>
gchristensen: Still need any?
drakonis has joined #nixos-chat
<andi->
cransom: are you liking it?
drakonis_ has quit [Ping timeout: 252 seconds]
<cransom>
andi-: that's really hard to quantify. it is stretching my brain, which is good. the part where it deviates from most every applications shortcuts isn't that much fun.
<andi->
cransom: those shortcuts are what keeps me from switching to anything but qwerty..
<cransom>
it has been interesting learning how different tactile responses (feel of a different board) automatically reprogram my fingers for a different layout.
tilpner has quit [Ping timeout: 245 seconds]
<cransom>
so switching back and forth isn't a big deal.
<gchristensen>
1
<gchristensen>
+1
<cransom>
I had a neo matrix moment when I assigned arrow keys to the a modifier so the same way I move around in vim (was hjkl), I can do that every where now.
<__monty__>
Hmm, weird. I had no problem whatsoever with keybindings. Only annoyance is wasd games that don't have reprogrammable controls.
<__monty__>
I even prefer the weird hjkl position for vim.
<cransom>
control/command x z c v aren't convenient to find on one hand without concious effort, for me at least.
<__monty__>
cransom: One hand? What are you learning touch typing for? : >
<__monty__>
I switched in highschool though so I probably had an easier time overcoming muscle memory.
<cransom>
mice or other pointing devices usually require a hand off the key.v
<cransom>
*keys
<cransom>
I have mouse keys around too, but they aren't That convenient
<__monty__>
I've just learned to love the terminal. Mousing is restricted to browsing the web (unless I'm on a browser with good keyboard interaction).
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 252 seconds]
<mdash>
honestly I wonder how much less off-putting people would find nix syntax if it used the same syntax as JSON for lists and maps
<__monty__>
I think the offputting bit is how minimal it is. It feels like things are trying to sneak past you.
tilpner has joined #nixos-chat
drakonis has joined #nixos-chat
<infinisil>
mdash: I think syntax is the least grave problem nix has
<gchristensen>
what is most grave?
drakonis_ has quit [Ping timeout: 255 seconds]
<andi->
i've heard many times that the syntax was the thing that turned people off.. Not sure what they'd have rather tho..
<infinisil>
gchristensen: I think the privacy thing is a really bad one
<gchristensen>
privacy?
<infinisil>
The fact that there is none for the store
<andi->
you mean a way to store "private" secrets?
<infinisil>
Then maybe another really bad one is the fact that it's almost impossible to debug nix
<gchristensen>
a step-thru debugger would be so nice
<infinisil>
Yeah
__monty__ has quit [Quit: leaving]
<infinisil>
And then another one might be that on a modern machine with GBs of RAM and nGHz processors it still takes multiple *seconds* to `nixos-rebuild --switch`
<gchristensen>
we pay a heavy tax for cross compilation :(
<infinisil>
Do we?
<gchristensen>
yes
<infinisil>
How do you know?
<gchristensen>
eval time & ram went up substantially after cross
<infinisil>
Ahh
<infinisil>
I also think the number of options is a problem
<LnL>
yeah, it was pretty noticeable with a large evaluation and it has not gotten better since then
<LnL>
but for nixos specifically it's mostly the modules that add a lot of overhead
yl has quit [Ping timeout: 246 seconds]
<infinisil>
So the reason these options are evaluated (all of them) is because the configuration.nix man page needs it
<infinisil>
Simple idea that might solve this: nixos-rebuild should do 2 nix-build's in parallel, one for the system, one for the man pages
<LnL>
oh right, what was that again?
<infinisil>
Then link them together in a final step
yl has joined #nixos-chat
<infinisil>
I might just give this idea a go
<infinisil>
LnL: What was what again?
<andi->
that only makes sense if you have enough CPU & RAM to run both at the same time... Sometimes a VM with 2GB is already having trouble :/
<infinisil>
andi-: Yeah, so we could add switches for turning off this parallel build, and switches for *only* doing the system, not rebuilding the man page
<infinisil>
That would actually be very nice
<LnL>
the evaluation issue, I'm pretty sure what's needed for the manpages can be moved inside the build
<infinisil>
LnL: error while parsing sentence
<infinisil>
Oh but problem: The system derivation wants the man pages in its input. replaceDependency ftw lol
<LnL>
if the nixos manpage takes <nixpkgs/nixos> as input and evaluates the options -> json or whatever at build time it's not necessary anymore for the drv
<infinisil>
Hmm interesting idea, I'm not sure if this usage of recursive nix will work though
<infinisil>
With all that not-being-able-to-write-to-the-store business
<LnL>
eval doesn't need to write to the store, nix-instantiate actually doesn't by default unless you pass --read-write-mode IIRC
<infinisil>
Yeah, but I think there's a bunch of things that still do require read-write-mode even though they technically shouldn't, I think I've seen some issue regarding that at some point
<infinisil>
But that's worth trying, I'll probably give it a shot
<LnL>
but that could be replaced with nix-instantiate --xml > options.xml
<infinisil>
Hmm yeah
<LnL>
the channel builds do something similar, to build a package index
ar1a has joined #nixos-chat
<andi->
all those altcoin software that we have in nixpkgs /o\ it feels like a dumbster. Am I the only one that dislikes that whole stuff?
* samueldr
wonders if dumBster was intentional
<ar1a>
i think crypto has a lot of potential, ONCE its used as a currency
<ar1a>
and not as a money making scheme
<andi->
samueldr: ;)
<simpson>
andi-: It's pretty funny. I appreciate that Nix can be used to refactor any kind of disastrous build system~
<samueldr>
I'm not too keen of the altcoins attrset; not sure how maintained it is; and are they all still alive?
<andi->
I think it is funny that there were breaking "mandatory" updates to some of those coins and our packages haven't been touched... Means I can remove it?
<samueldr>
though, not yucking anyone's yum, if they are, and are used, then good for them users
<samueldr>
andi-: smells like it to me :/
<andi->
I am also tempted to just ignore them on my list of openssl1.1 packages that currently fail..
joepie91 has quit [Ping timeout: 252 seconds]
<infinisil>
LnL: Unfortunately the building options.json (and xml) from a path to nixpkgs doesn't work well
<infinisil>
Problem: We need pkgs to evaluate the docs, for related packages and for valid attr names, so we need to import <nixpkgs> into the store, which could be wayyyy too slow
<infinisil>
NixOS allows users to change the baseModules by passing an *expression*. We would have to somehow serialize that expression so the recursive Nix can read it, or alternatively deprecate support for that
<infinisil>
(^ that is the second problem)
<LnL>
:/
<infinisil>
Third problem: Any additional user modules have to chance to get docs (which would now be supported since a recent PR)
<infinisil>
s/to/no
yl has quit [Ping timeout: 268 seconds]
<infinisil>
So I think I'll try out my parallel manual thing instead next
jackdk has joined #nixos-chat
<gchristensen>
andi-: do it
<gchristensen>
andi-: ignore them.
<gchristensen>
andi-: if they don't get fixed in a bit, we can EASILY delete them ;)
<pie_>
andi-, could be ocasionally useful if there was some documented way to search for any packages that ever existed if someone comes back later and tries to update smething? :V
<andi->
pie_: git log? :D
<pie_>
i shoudl stop wanting obscure nice things
<pie_>
andi-, would that actually work
<andi->
the github search should be sufficient or any other code platform search that we might be using in the future